Run experiments using tada-hdt-entity library
- T2Dv2
classes_with_col_GS.csv
: this file contains the file names, the column id of the subject columns, and their classes This is update version from theclasses_GS.csv
provided by the T2Dv2. We changed the concept column with the subject column id and we fixed some mislabels.classes_with_col_GS_2016_04
: a subset ofclasses_with_col_GS.csv
, which contains the entity columns which exists in dbpedia's hdt of 04-2016.properties_entity.csv
: contains the properties with entities as the objectsproperties_2016_04.csv
: a subset ofproperties_entity.csv
which only contains the properties which exists in dbpedia's hdt of 04-2016.tranformer.py
: a python script to generateproperties_entity.csv
from the annotation files of T2Dv2.
- To generate
properties_entity.csv
.python tranformer.py "T2Dv2_dir/property" classes_with_col_GS_2016_04.csv properties_entity.csv
You can replaceT2Dv2_dir
with the directory containing the T2Dv2 folder. You can download it from here. - To generate
properties_2016_04.csv
, you need to change themain.cpp
and include the following source code:
#include<iostream>
#include<string>
#include<stdio.h>
#include "t2dv2.h"
#include <tabular_parser/parser.h>
using namespace std;
int main(int argc, char* argv[]) {
T2Dv2* t2d = new T2Dv2(hdt_file, log_file, classes_file, files_dir);
t2d->set_lang_tag("@en");
t2d->set_title_case(false);
t2d->generate_properties_file("datasets/t2dv2/properties_entity.csv", "datasets/t2dv2/properties_2016_04.csv");
return 0;
}
and then compile and run make;bin/tadaeexperiment